Reply length
Automatic adds nothing to the prompt and lets the model decide. The other three append a length instruction. The model still matches your intent: a greeting gets a short reply even on Long.
Individual profiles can override this, and the profile’s choice wins for its own turns.
Conversation memory
How many recent messages are sent as context each turn. Higher keeps more of the thread in mind, at the cost of a bigger prompt. Set it to 0 to treat every message as standalone.
Summarize long chats
On by default. When the conversation grows past the messages-to-remember window, the oldest messages are folded into a rolling summary instead of being dropped. The most recent messages stay verbatim. The summary is injected ahead of the history, so a long chat keeps its beginning in condensed form. Turn it off and the oldest messages are simply lost once they fall outside the window.What triggers a summary
What triggers a summary
The trigger is the conversation memory setting above, not the model’s
context window. The fold runs in the background after a reply finishes, so
it never delays an answer. If it fails, the summary is left alone and the
next turn tries again.
Profiles
Give the assistant different personas, each with its own reply length.
Personal memory
Durable facts the assistant remembers across chats.